Despite major advances in cardiovascular surgical techniques and the routine use of prophylactic antimicrobial agents, prosthetic valve endocarditis (PVE) continues to complicate the course of a small percentage of patients after cardiac valve replacement. The prognosis of PVE is grim, especially when cardiac and extracardiac complications occur.1,2 One of the most dreaded complications arising from PVE is congestive heart failure (CHF), which is the most frequent indication for early surgery3 and has been identified as an independent risk factor for early and late mortality in patients with native valve endocarditis4 and PVE.2,5,6

There is scant data in past literature that specifically addresses the significance of CHF in PVE. The importance of CHF as a prognostic factor and the identification of the best therapeutic approach towards its treatment are key issues for clinicians involved in the management of patients diagnosed with this challenging condition.

This study represents the widest series to date with the largest study population, which specifically analyzes the incidence and causes of CHF in patients with PVE. The study describes the profile of CHF in these patients, identifies prognostic factors associated with its development, addresses its prognostic significance, and analyzes clinical outcomes based on chosen therapeutic strategies.

Patients included in the analysis were enrolled from 3 university-affiliated tertiary care hospitals; these were the referral centers for their regions on infective endocarditis. All the hospitals worked together using standardized protocols, uniform data collection, and uniform diagnostic and therapeutic criteria from the beginning of the study.

From 1996 to 2009, 639 episodes of definite left-sided infective endocarditis in 619 patients were prospectively enrolled for the study; Duke criteria were applied until the year 200211 and modified Duke criteria were applied thereafter.12 Of these 639 episodes, 257 (40%) were PVE and constituted the study group.

Among 257 patients with PVE in our series, 145 patients (56%) were diagnosed of CHF: 115 patients (79%) had CHF at the time of admission, (58% patients from NYHA functional class III or IV) and the remaining 30 patients (21%) developed CHF during their hospital stay.

Despite the fact that the development of CHF in patients with PVE is a crucial aspect in the prognosis of these patients, there is a lack of studies in previous literature that specifically address this issue. Very important aspects of this entity, as its incidence and its influence on the prognosis or predictors of mortality, remain unknown. This study, that represents the widest series to date which specifically analyzes CHF in patients with PVE, has shed some light on these unknown aspects.

Several consequences can be drawn from this study. First, the incidence of CHF among patients with PVE is very high, accounting for more than 50% of these patients. Second, the mortality of PVE complicated with CHF is very high, especially in patients who receive medical treatment alone. Third, CHF is an independent risk factor associated with PVE, which increases the risk of in-hospital mortality in patients by threefold. And fourth, persistent infection, aortic involvement, abscess, and diabetes mellitus are the independent risk factors associated with mortality in patients with PVE and CHF. On the other hand, cardiac surgery is shown to decrease mortality in patients with PVE and CHF.

To the best of our knowledge, the incidence of CHF is the highest in our series of patients with PVE. Several reasons might explain this finding: expert teams on heart failure were responsible for these patients and these teams followed the most universally accepted criteria for the diagnosis of CHF-the Framingham criteria. Also, the aggressive microbiological profile of our series, with a high proportion of infections caused by staphylococci species may explain the high incidence of CHF in this study. Finally, in contrast to other studies, we included patients with CHF from all NYHA functional classes, not just NYHA functional classes III and IV.13

Our results suggest that cardiac surgery decreases mortality in patients with PVE and CHF. This finding must be interpreted with some caution, as surgery is very often denied in these patients because of the high operative risk involved in cardiac surgery and also because this subgroup of patients has the worst prognosis.14 This phenomenon is probably more frequent in PVE than in native valve endocarditis, as the patients are usually older in PVE cases and have at least 1 previous cardiac intervention. Only randomized studies can demonstrate whether early surgery actually improves the prognosis of these patients.15 Nonetheless, our data reinforces what the guidelines recommend; that the best therapeutic option for patients with PVE and CHF is cardiac surgery.13–16

From a clinical point of view, the identification of high-risk subgroups of patients based on the presence of prognostic markers is very important in both native valve endocarditis and PVE, because it helps us to decide the best therapeutic approach for these patients. Several predictive factors of mortality in patients with PVE have been identified in previous studies17: increasing age, severe comorbidity, persistent bacteremia, health care associated infection, S. aureus, early onset PVE, renal failure, mediastinitis, complicated PVE, abscesses, cerebral complications, CHF, and septic shock. Among them, staphylococcal infections and CHF are the most constant variables. Our results corroborate these findings, and include diabetes mellitus as an important prognostic marker of in-hospital mortality in PVE. Some previous studies that have attempted to identify risk factors in patients with PVE are now outdated,18–20 as these studies analyzed only a specific subgroup of patients with PVE,18–21 included only a few patients,19,21 or came from single institutions.18–22 The international-collaboration-on-endocarditis group reported 556 patients with PVE from 63 centers in 28 countries.2 They reported older age, health care-associated infection, S. aureus, CHF, stroke, intracardiac abscesses, and persistent bacteremia as predictive factors of in-hospital mortality in patients with PVE. In their study, diabetes mellitus did not reach statistically significant differences and renal failure was not analyzed.

Once it was established that CHF worsens the prognosis of patients with PVE, we further investigated what factors help in identifying high-risk patients with PVE and CHF. The 4 independent risk factors identified in our series have already been described in previous studies.2,5,23,24 The most powerful predictor of mortality in our series was persisting infection, which was in consonance with previous findings of our group.25 It is important to highlight that more than 70% of the PVE cases caused by S. aureus were associated with persistent infection, which explains why this microorganism did not reach significance in the multivariate analysis. The role of perivalvular complications in the prognosis of infective endocarditis has been largely studied in detail and is well established.2,10 However, less uniformity in data is found regarding diabetes mellitus. In some series, diabetes mellitus was shown to increase the risk of mortality in patients with infective endocarditis,26 however, these results are not consistent with findings from other authors.27 Our results suggest that diabetes mellitus may be associated with increased risk when the infectious process affects prosthesis. Mechanisms proposed in previous literature relate to depression in the leukocyte chemotaxis, adherence, phagocytosis, intracellular killing, and opsonization.26

Finally, this study was done at tertiary hospitals, a setting that biases the type of patients included in our database. Our sample does not reflect the characteristics of all patients with infective endocarditis in the general population, but rather the population of patients with PVE admitted to dedicated hospitals. Accordingly, our conclusions are applicable to reference hospitals equipped and staffed to perform heart surgery. All observational studies, as well as many randomized studies, have built-in bias.29 Effects of referral bias in tertiary care centers have been previously acknowledged in endocarditis30 and in surgical outcomes.31 It cannot be avoided but it must be recognized. In our population of patients with PVE, mortality may be higher than that observed in a population-based cohort because patients may have died before they are sent to our hospital. On the contrary, mortality may be lower because patients with a favourable clinical course may not be sent to our hospital. However, our results are comparable to other series coming from tertiary centers, where majority of the research in infective endocarditis has been performed.

The present study was partly financed by the Cooperative Cardiovascular Disease Research Network and was supported by the Instituto de Salud Carlos III in Spain.

None declared.